Reducing microglial lipid load enhances β amyloid phagocytosis in an Alzheimer's disease mouse model

Abstract

Macrophages accumulate lipid droplets (LDs) under stress and inflammatory conditions. Despite the presence of LD-loaded macrophages in many tissues, including the brain, their contribution to neurodegenerative disorders remains elusive. This study investigated the role of lipid metabolism in Alzheimer's disease (AD) by assessing the contribution of LD-loaded brain macrophages, including microglia and border-associated macrophages (BAMs), in an AD mouse model. Particularly, BAMs and activated CD11c⁺ microglia localized near β amyloid (Aβ) plaques exhibited a pronounced lipid-associated gene signature and a high LD load. Having observed that elevated intracellular LD content correlated inversely with microglial phagocytic activities, we subsequently inhibited LD formation specifically in CX3CR1⁺ brain macrophages using an inducible APP-KI/Fit2^iΔMφ transgenic mouse model. We demonstrated that reducing LD content in microglia and CX3CR1⁺ BAMs remarkably improved their phagocytic ability. Furthermore, lowering microglial LDs consistently enhanced their efferocytosis capacities and notably reduced Aβ deposition in the brain parenchyma. Therefore, mitigating LD accumulation in brain macrophages provides perspectives for AD treatment.

Fig. 1.. AD accelerates LD accumulation in brain macrophages.

(A) Schematic diagram showing the experimental design (image created with BioRender.com). (B) Representative confocal images showing DAPI (blue) and BODIPY (green) staining of sorted microglia from 2-month-old (2m) WT and APP-KI mice fed ND, 6-month-old (6m) WT and APP-KI mice fed ND, and 6-month-old WT and APP-KI mice fed HFD. Scale bars, 5 μm. (C and D) Bar chart (with individual values shown as dots) of the percentage of LD⁺ microglia and the LD area per microglia. Each dot represents the average result calculated from one slide; three to five slides were evaluated per mouse and n = 5 mice per group. (E) Representative confocal images showing DAPI (blue) and BODIPY (green) staining of sorted BAMs. Scale bars, 5 μm. (F and G) Bar chart (with individual values shown as dots) of the percentage of LD⁺ macrophage and LD area per macrophage. Each dot represents the average result calculated from one slide; three to five slides were evaluated per two to three pooled mice and n = 6 mice per group. (H) Representative images showing anti-Iba antibody (green), anti-Aβ antibody (white), LipidSpot 610 (red), and DAPI (blue) staining of the brain tissue from a 6-month-old HFD-fed APP-KI mouse. An area with a radius of <35 μm from the center of the plaque was defined as the plaque-proximal area. Scale bars, 20 and 5 μm (zoom-in). (I) Quantification of LD⁺ microglia inside and outside the plaque-proximal area. (J) Representative confocal images showing CD206 (green), LipidSpot 610 (red), and DAPI (blue) staining of the brain border structures (dura mater and subdural meninges) from a 6-month-old HFD-fed APP-KI mouse. Scale bars, 10 and 2 μm (zoom-in). Samples were analyzed by two-way ANOVA. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. For clarity, nonsignificant values are not shown.

Fig. 2.. Lipidomic signature shift in microglia from APP-KI mice.

(A) Schematic diagram showing the experimental design (image created with BioRender.com). (B) 2D PCA plots depicting distinct lipid metabolite profiles of microglia isolated from 6-month-old HFD-fed WT and APP-KI mice. For each data point, sorted microglia (2 × 10⁵ cells) were obtained from a pool of two to three mouse brains. (C) Bar chart showing the percentages of distinct lipid categories detected in microglia isolated from WT and APP-KI mice. Lipid categories were classified according to the LIPID MAPS Structure Database (www.lipidmaps.org/databases/lmsd/browse). FA, fatty acyl; DG, diglyceride; MG, monoglyceride; Cer, ceramide; HexCer, hexosylceramide; SM, sphingomyelin; LPC, lysophosphatidylcholine; PA, phosphatidic acid; PC, phosphatidylcholine; PI, phosphatidylinositol; PS, phosphatidylserine. (D) Hierarchical clustering heatmap revealing major alterations in lipid metabolites in microglia from WT and APP-KI mice. Samples were analyzed by two-way ANOVA. *P < 0.05; **P < 0.01. For clarity, nonsignificant values are not shown.

Fig. 3.. scRNA-seq reveals a pronounced lipid signature in the CD11c-expressing microglia.

(A) Schematic diagram showing the experimental design (image created with BioRender.com). Parenchyma and brain borders were processed separately, and obtained microglia and BAMs were sorted and pooled in a ratio of 9 to 1. For each experimental group, sorted microglia and BAMs were obtained from a pool of three mouse brains. (B) UMAP plot showing all single-cell data obtained from 6-month-old HFD-fed WT and APP-KI mice that passed quality control. HM1-3, homeostatic microglia 1-3; ARM (Itgax^hi), activated-response microglia; TRM, transiting-response microglia; IRM, interferon-response microglia; Nlrp3^hi, Nlrp3^hi microglia; Gpr84^hi, Gpr84^hi microglia; Tnf^hi, Tnf^hi microglia; Trem2^neg, Trem2-negative microglia. (C) Percentage of each cell cluster identified in the WT and APP-KI groups of mice. (D) UMAP plots displaying the average expression levels of the top 30 genes expressed by the foam cell cluster identified in (5) in WT and APP-KI mice. (E) Violin plots showing the expression levels of genes related to lipid metabolism in each cluster. (F) Representative confocal images showing DAPI (blue) and BODIPY (green) staining of sorted CD11c⁺ and CD11c⁻ microglia from 6-month-old HFD-fed APP-KI mice. Scale bars, 2 μm. (G) Bar chart (with individual values shown as dots) of the LD areas of CD11c⁻ microglia and CD11c⁺ microglia. Each dot represents the average result calculated from one slide; three to five slides were evaluated per mouse with n = 3 mice per group. Samples were analyzed by Student’s t test. **P < 0.01.

Fig. 4.. FIT2-deficient microglia and CD206⁺ BAMs accumulate fewer LDs.

(A) Schematic layout showing the experimental design (image created with BioRender.com). (B) Representative confocal images of bulk microglia isolated from 6-month-old HFD-fed TAM-treated and untreated APP-KI/Fit2^iΔMφ mice stained with BODIPY (green) and DAPI (blue). Scale bars, 5 μm. (C) Quantification of the LD area in microglia from TAM-treated and untreated APP-KI/Fit2^iΔMφ mice. Each dot represents the average result calculated from one slide; three to five slides were evaluated per mouse with n = 4 mice per group. (D) Representative images showing DAPI (blue), anti-Iba antibody (red), and LipidSpot 610 (green) staining of brain tissue sections from 6-month-old HFD-fed TAM-treated and untreated APP-KI/Fit2^iΔMφ mice. Scale bars, 10 μm. (E) Bar charts (with individual values shown as dots) of the percentage of LD⁺ microglia obtained from TAM-treated and untreated APP-KI/Fit2^iΔMφ mice. Each dot represents the average result calculated from one slide; three to five slides were evaluated per mouse with n = 5 mice per group. (F) Representative images showing CD206⁺ BAMs isolated from 6-month-old HFD-fed TAM-treated and untreated APP-KI/Fit2^iΔMφ mice, stained with DAPI (blue) and BODIPY (green). Scale bars, 5 μm. (G) Quantification of the LD⁺ area in CD206⁺ BAMs obtained from TAM-treated and untreated APP-KI/Fit2^iΔMφ mice. Each dot represents the average result calculated from one slide; three to five slides were evaluated from a sample of three pooled mice with n = 6 mice per group. Samples were analyzed using Student’s t test. **P < 0.01.

Fig. 5.. FIT2 depletion alters the lipidomic and transcriptomic profiles of microglia in the AD brain.

(A) 3D PCA plot showing the distinct lipid metabolite profiles of microglia isolated from 6-month-old HFD-fed TAM-treated and untreated APP-KI/Fit2^iΔMφ mice. For each data point, sorted microglia (2 × 10⁵ cells) were obtained from a pool of two to three mouse brains. (B) Hierarchical clustering heatmap showing major alterations in lipid metabolites in microglia from TAM-treated and untreated groups. (C) Bar chart showing changes in the levels of different TGs among the microglia isolated from 6-month-old HFD-fed WT, APP-KI, and APP-KI FIT2^−/− mice. (D) Violin plots showing the expression levels of lipid metabolism–related genes in CD11c⁺ microglia isolated from TAM-treated and untreated mice. Microglia were sorted from a pool of three mouse brains for each experimental group, as indicated in the figure legend. (E) Volcano plot displaying the distinct gene expression profiles of CD11c⁺ microglia isolated from the TAM-treated and untreated mice. (F) Flow cytometry histograms showing the MHCII expression levels of CD11c⁺ microglia isolated from TAM-treated and untreated mice. (G) Bar chart (with individual values shown as dots) showing the mean fluorescence intensity (MFI) of MHCII. Each dot represents a value obtained from one mouse brain and n = 6 mice per group. (H) Gene set enrichment analysis of lysosomal and phagosomal pathways of CD11c⁺ microglia isolated from TAM-treated and untreated mice. Samples were analyzed using Student’s t test. *P < 0.05.

Fig. 6.. FIT2 deficiency increases the phagocytic activity of microglia and BAMs.

(A) Schematic layout showing the experimental design (image created with BioRender.com). AF647, Alexa Fluor 647. (B) Representative images showing microglia isolated from 6-month-old HFD-fed TAM-treated and untreated APP-KI/Fit2^iΔMφ mice stained with LipidSpot 610 (red), DAPI (blue), and engulfment of E. coli bioparticle (green). Scale bars, 10 μm. (C) Bar chart of the percentages of E. coli⁺ microglia in TAM-treated and untreated APP-KI/Fit2^iΔMφ mice. Each dot represents the average result calculated from one region of interest; five to seven regions of interest were evaluated per mouse with n = 5 mice per group. (D and E) Flow cytometry plots and bar chart displaying the percentages of E. coli⁺ CD11c⁻ and CD11c⁺ microglia isolated from TAM-treated and untreated APP-KI/Fit2^iΔMφ mice. Each dot in the bar chart represents the result obtained from one mouse and n = 8 mice per group. (F and G) Flow cytometry plots and bar chart displaying the percentages of Aβ oligomer⁺ CD11c⁻ and CD11c⁺ microglia isolated from TAM-treated and untreated APP-KI/Fit2^iΔMφ mice. Each dot in the bar chart represents the result obtained from one mouse and n = 7 mice per group. (H and I) Flow cytometry plots and bar graph showing the percentages of E. coli⁺ MHCII^hi/int BAMs and MHCII^lo BAMs isolated from TAM-treated and untreated mice. Each dot represents the percentage of E. coli⁺ cells obtained from one mouse brain and n = 8 mice per group. (J and K) Flow cytometry plots and bar graph showing the percentages of Aβ oligomer⁺ MHCII^hi/int BAMs and MHCII^lo BAMs isolated from TAM-treated and untreated mice. Each dot in the bar chart represents the percentage of Aβ⁺ cells obtained from one mouse brain and n = 5 to 6 mice per group. Samples were analyzed by Student’s t test (C) and two-way ANOVA [(E), (G), (I), and (K)]. *P < 0.05; **P < 0.01. For clarity, nonsignificant values are not shown.

Fig. 7.. Microglial FIT2 deficiency reduces Aβ load in the AD brain.

(A) Diagram illustrating the use of methoxy-XO4 in the in vivo phagocytosis assay (image created with BioRender.com). ip, intraperitoneally. (B) Representative flow cytometry plots displaying the methoxy-XO4⁺ CD11c⁻ and CD11c⁺ microglia isolated from 6-month-old HFD-fed TAM-treated and untreated APP-KI/Fit2^iΔMφ mice. (C) Bar chart (with individual values shown as dots) of the percentage of methoxy-XO4⁺ CD11c⁻ and CD11c⁺ microglia isolated from TAM-treated and untreated mice. Each dot represents the result obtained from one mouse brain and n = 8 mice per group. (D) Representative images showing anti-Aβ antibody (82E1) staining of the brain tissues of TAM-treated and untreated APP-KI/Fit2^iΔMφ mice. Scale bar, 100 μm. Blue outline, cortex; red outline, hippocampus. (E) Upper panel: Representative confocal images of anti-Aβ (blue) and anti-Iba (red) antibody staining of brain tissues. Scale bars, 10 μm. Lower panel: Representative confocal images of ThioS (green) and anti-Iba antibody (red) staining of brain tissues. Scale bars, 10 μm. (F) Quantification of the anti-Aβ antibody–positive plaques in the cortex and hippocampus. Each dot represents the average result obtained from an individual mouse brain with n = 5 mice per group. (G) Quantification of the ThioS-stained Aβ plaques in the cortex and hippocampus. Each dot represents the average result obtained from an individual mouse brain with n = 4 to 5 mice per group. Samples were analyzed by two-way ANOVA (C) and Student’s t test [(F) and (G)]. *P < 0.05; **P < 0.01.